For a discussion in another thread I made some comparisons between a single driver, an array of 25 drivers and an array of
25 drivers and their floor/ceiling mirrors. Using the dimensions of my room, meaning the gaps that are present between the
lowest driver at the flor and it's mirror and ditto for the ceiling.
As it's about my project and it hasn't been shown on this thread before, I figured I should add it here.
For my own interest I've also made the same comparison using my filtered array:
What's clearly seen between these two sims is the reduction in comb filtering on the top end of the filtered array. You
can see how the original frequency shape stays in-tact even on the top end. The 3dB drop per octave is extended up to
10 KHz. The only thing keeping it from being a perfect array is the lobing on top creating the reflections off of the floor
and ceiling. Believe me, I've been over this hundreds of times in my head of how I could solve this without ruining
anything else (like the looks). We've shown that a sort of waveguide shape between drivers "could" help, but how to
keep it looking like it belongs there. It's not unthinkable to go for a segmented total waveguide steering the horizontal
spread as well, but how large would it need to be? Would we still get pattern flip etc.
Incremental steps have gotten me this far... who knows what the future brings
.
25 drivers and their floor/ceiling mirrors. Using the dimensions of my room, meaning the gaps that are present between the
lowest driver at the flor and it's mirror and ditto for the ceiling.
As it's about my project and it hasn't been shown on this thread before, I figured I should add it here.
For my own interest I've also made the same comparison using my filtered array:
What's clearly seen between these two sims is the reduction in comb filtering on the top end of the filtered array. You
can see how the original frequency shape stays in-tact even on the top end. The 3dB drop per octave is extended up to
10 KHz. The only thing keeping it from being a perfect array is the lobing on top creating the reflections off of the floor
and ceiling. Believe me, I've been over this hundreds of times in my head of how I could solve this without ruining
anything else (like the looks). We've shown that a sort of waveguide shape between drivers "could" help, but how to
keep it looking like it belongs there. It's not unthinkable to go for a segmented total waveguide steering the horizontal
spread as well, but how large would it need to be? Would we still get pattern flip etc.
Incremental steps have gotten me this far... who knows what the future brings
.
If I up the level per driver to match the single TC9, things become more clear why this array business is attractive:
Now we see that the SPL level in total is 10 dB higher everywhere compared to the single driver. 14.14volts applied to the array.
The same 1 watt per driver as in the single driver plot, so a fair comparison and proof there is headroom available for EQ.
The gray in-room prediction is at 2.7 meter listening distance (in both cases). No EQ!
Now we see that the SPL level in total is 10 dB higher everywhere compared to the single driver. 14.14volts applied to the array.
The same 1 watt per driver as in the single driver plot, so a fair comparison and proof there is headroom available for EQ.
The gray in-room prediction is at 2.7 meter listening distance (in both cases). No EQ!
Last edited:
A quick note: in this comparison there is about 10 dB of boost on the low end! I forgot I had selected EQ (I even noted: no EQ)
to counter the single driver frequency changes. Here it is without EQ:
Its nice you make the two switch by creating an animated gif, but its quite annoying to look at when you want to look at it in detail. Also its not directly clear which version you look atFor a discussion in another thread I made some comparisons between a single driver, an array of 25 drivers and an array of
25 drivers and their floor/ceiling mirrors. Using the dimensions of my room, meaning the gaps that are present between the
lowest driver at the flor and it's mirror and ditto for the ceiling.
As it's about my project and it hasn't been shown on this thread before, I figured I should add it here.
View attachment 1078617
For my own interest I've also made the same comparison using my filtered array:
View attachment 1078618
What's clearly seen between these two sims is the reduction in comb filtering on the top end of the filtered array. You
can see how the original frequency shape stays in-tact even on the top end. The 3dB drop per octave is extended up to
10 KHz. The only thing keeping it from being a perfect array is the lobing on top creating the reflections off of the floor
and ceiling. Believe me, I've been over this hundreds of times in my head of how I could solve this without ruining
anything else (like the looks). We've shown that a sort of waveguide shape between drivers "could" help, but how to
keep it looking like it belongs there. It's not unthinkable to go for a segmented total waveguide steering the horizontal
spread as well, but how large would it need to be? Would we still get pattern flip etc.
Incremental steps have gotten me this far... who knows what the future brings
the 10dB I don't understand. With 25 drivers you would expect 20log10(25)=27.9dB gain for low (in phase) frequencies.
What am I understanding wrong?
I can only agree, I figured you'd see the progression of it, but that didn't really happen.Its nice you make the two switch by creating an animated gif, but its quite annoying to look at when you want to look at it in detail. Also its not directly clear which version you look at
Did you check at ~200 to 250 Hz?
The 10 dB I mentioned is where there isn't an in-phase addition, as in the top end. Let me explain why I said that. Obviously, over the years I've read a lot of topics on line arrays. In those discussions, lots of things are being said. I'm the type of guy that remembers stuff he finds interesting. So once upon a time, a member here (yes I do remember the name) made a remark that went something like this: "these arrays sum up quite nice at low frequencies, but their top end isn't any more efficient than that of a single driver."
(not quoting the actual thing, as that isn't important)
While one can think that if you're looking at these two pictures side by side:
At 1 watt into the single driver or the array, we have a 13 dB increase at 200 to 300 Hz. Just look at the light grey prediction. Which is the prediction
at 2.7 m listening distance in-room. You could follow the black line in this instance too, but as I was also adding the reflections into the picture, the
light gray prediction is the one to "see". At 200 Hz, we go from 71 dB to 84 dB. At 300 Hz we go from 73.5 to 86.5 dB. Both a net gain of 13 dB, right?
Above it, we get the ~3 dB drop per octave. This part is discussed in the line array theory thread.
This won't be exactly right, due to the driver distance and specific driver directivity etc. as we're not talking about an ideal line source here, just an
approximation. Yet, the trend is visible. At just over 4 KHz, we see the array and the single driver having the same output level. Remember that line
I wrote, at the start of this post? "these arrays sum up quite nice at low frequencies, but their top end isn't any more efficient than that of a single driver."
Which is true, as with an input of 1 watt into the speaker, the array becomes less efficient and even drops below the single driver output due to combing
and loss of directivity. It's also the end of the 3 dB loss per octave, but substituted for that we see the effects of combing etc. On average it is more flat
than the previous frequency curve, meaning it doesn't quite follow the array trend like it did for the previous section. Up until about 10 KHz we are
save from large nulls due to combing, which is why I believe these effects are not annoying to listen to. If we take care in processing and don't let our
automated processing try to restore any large peaks above 10 KHz it will sound quite well!
I've mentioned it often, here and there, that I always go over the processing result and reduce over processing where needed. Well, that's where it's
needed. If you don't reduce it, it won't be that obvious immediately, but moving slightly off axis might result in similar looking narrow peaks. Those
could become audible, much more so than the narrow dips we see in this second picture.
Anyway, we had a third picture, one where we added the mirror images of the floor and ceiling:
While this may look horrible, it's not how we "perceive" it. We see the top end becoming wild and fuzzy, but what we perceive will be closer to a
filtered version of it, yet the truth is that it looks like this when measured in the room.
I can show the single driver + similar room effects for a reminder:
That's the equivalent of the single driver plus room effects. That's not how we "perceive" a single driver in our room, right? Yet we all have seen
how measurement change when moving about in the room and we use either gating or smoothing tools to make sense of our curves. one of those
smoothing tools is the "Psy" filter. which is designed to "come close" to how we perceive this effect. Apply it in your thoughts as I can't be bothered
to do it right now
.But the choice here is: do we want a good looking frequency curve below 4K and a fuzzy top end or do we prefer a fuzzy everything below 8K?
(while keeping in mind what our brain does with these sounds)
Those two pictures are "kind of" the differences between perceived sound of a single driver vs an array. Well, there are other factors at play but
if you have heard a reasonably well thought out array, corrected in a room... and with reasonably well thought out, I do mean that there was care
taken with the parallel planes and ridges that I always bring up in these discussions... then you may understand where that difference in perception
comes from looking at these pictures...
I think I'll press "post reply" now, and continue in a new post...
And all is well again
One other thing we looked at in the other thread, is if there was an increase in output due to having the floor and ceiling reflections make the array seem longer than it is,
First the unfiltered array:
Next the same array but with floor and ceiling mirrors:
As said in the other thread: the gap between the array and the ceiling may be too large to work continuously, the gain is minimal.
4 dB at 20 Hz is what we see here and it's basically done by 100 Hz, no more increase. There will be some room gain in a small enough
room and if the array is close enough to a back wall you'll see some boundary gain from that. And lets be honest, in that area it needs all
the help it can get, so place it close to the back wall! What has worked best for me is the baffle 0.5 meter from the back wall. This gain
is variable with position, a corner probably works best, yet that's the biggest enemy of the array in that the side wall is a parallel plane...
It will also have spots where it doesn't do as well. Anyway... I've been over that in this thread a million times.
Bottom line: if there is free energy to be had: use it! That's why even the bottom end of the array still helps out to get the frequency curve
I want to have, but only where the room is helping the array to achieve it. So that these silly small drivers don't have to work so hard to
achieve it. That has brought about some controversy in the past of this thread on more than one occasion....
Yet I believe in using this "nearly free" energy to have more low frequency sources to balance out the low end in the room. No, I don't use
the arrays where they don't get that kind of support. I look at distortion plots and determine where they do the easy lifting and use that to
my advantage. So no abrupt cut-off with a crossover at ~160 Hz, as the arrays are still loafing around till 60 Hz and if the room permits,
even lower than that. After all, it still does 85 dB at 1 watt (per driver) at 60 Hz(*). I let the distortion tell me where it does it job easily. If it
fails at any frequency below 70 Hz, back off the gas at that spot and let another component take over. That used to be the other array, but
now I can lift it up with 2 subwoofers and an array. That makes it possible to keep it stereo up until to that last octave where the subs do
all the lifting. There is headroom in the array before we even run out of steam with each driver. Throwing that away if you don't have the
option to use a couple of 18" subwoofers would be silly. So yes, I do still use boost, but with one eye on the distortion graphs.
(*) = at 2.7 meter or the listening spot. The array is doing 95 dB/at 1 watt per driver at 60 Hz, that's the loafing around. In comparison
the single driver at 1 watt is doing 68 dB, there's the 27 dB of gain again.
** A while ago, after I revealed my devious plan on why I had some feet bench beside the arrays (from the start of this project) I had to
promise: this is it. 2 arrays, 2 ambient speakers and 2 subwoofers. No more drivers! No more amplifiers! I intent to keep that promise**
One other thing we looked at in the other thread, is if there was an increase in output due to having the floor and ceiling reflections make the array seem longer than it is,
First the unfiltered array:
Next the same array but with floor and ceiling mirrors:
As said in the other thread: the gap between the array and the ceiling may be too large to work continuously, the gain is minimal.
4 dB at 20 Hz is what we see here and it's basically done by 100 Hz, no more increase. There will be some room gain in a small enough
room and if the array is close enough to a back wall you'll see some boundary gain from that. And lets be honest, in that area it needs all
the help it can get, so place it close to the back wall! What has worked best for me is the baffle 0.5 meter from the back wall. This gain
is variable with position, a corner probably works best, yet that's the biggest enemy of the array in that the side wall is a parallel plane...
It will also have spots where it doesn't do as well. Anyway... I've been over that in this thread a million times.
Bottom line: if there is free energy to be had: use it! That's why even the bottom end of the array still helps out to get the frequency curve
I want to have, but only where the room is helping the array to achieve it. So that these silly small drivers don't have to work so hard to
achieve it. That has brought about some controversy in the past of this thread on more than one occasion....
Yet I believe in using this "nearly free" energy to have more low frequency sources to balance out the low end in the room. No, I don't use
the arrays where they don't get that kind of support. I look at distortion plots and determine where they do the easy lifting and use that to
my advantage. So no abrupt cut-off with a crossover at ~160 Hz, as the arrays are still loafing around till 60 Hz and if the room permits,
even lower than that. After all, it still does 85 dB at 1 watt (per driver) at 60 Hz(*). I let the distortion tell me where it does it job easily. If it
fails at any frequency below 70 Hz, back off the gas at that spot and let another component take over. That used to be the other array, but
now I can lift it up with 2 subwoofers and an array. That makes it possible to keep it stereo up until to that last octave where the subs do
all the lifting. There is headroom in the array before we even run out of steam with each driver. Throwing that away if you don't have the
option to use a couple of 18" subwoofers would be silly. So yes, I do still use boost, but with one eye on the distortion graphs
.(*) = at 2.7 meter or the listening spot. The array is doing 95 dB/at 1 watt per driver at 60 Hz, that's the loafing around. In comparison
the single driver at 1 watt is doing 68 dB, there's the 27 dB of gain again.
** A while ago, after I revealed my devious plan on why I had some feet bench beside the arrays (from the start of this project) I had to
promise: this is it. 2 arrays, 2 ambient speakers and 2 subwoofers. No more drivers! No more amplifiers! I intent to keep that promise**
Last edited:
Anyway, to finish off this bit today I'll put up the single driver, the unfiltered array and the filtered array for reference...
You tell me what you see here...
And I owe it to you guys to show at least one more animated gif, as I know how much you guys love those!
I'm assuming here that this comparison does tell a story...
You tell me what you see here...
And I owe it to you guys to show at least one more animated gif, as I know how much you guys love those!
I'm assuming here that this comparison does tell a story...
Last edited:
As I'm on a roll today showing pictures and sims, I want to show a couple more pictures to explain some choices...
The filtered array with only the ceiling reflections showing...
And the new (more hairy) rug we got after remodeling
If I could have put a rug on the ceiling without my other half noticing it, I would have.
The filtered array with only the ceiling reflections showing...
And the new (more hairy) rug we got after remodeling
If I could have put a rug on the ceiling without my other half noticing it, I would have
.